Improved Calibration of Building Models using Approximate Bayesian Calibration and Neural Networks

Git repository supporting the manuscript titled: "Improved Calibration of Building Models using Approximate Bayesian
Calibration and Neural Networks"

The research touches on the application of Approximate Bayesian Calibration for building energy simulation calibrations for existing buildings using the Sequential Monte Carlo approach. It details the benefits of the method and includes a case study of a large, complex, existing retail building.

Note, running this repository requires installation of Energyplus version 9.5.0. Running this directly in Mybinder or similar, we recommend running notebooks directly within the 'notebooks' folder and skipping the parameter sampling notebook stage. The sensitivity analysis and calibration will function with the included pre-sampled set.